MATERIAL Y MÉTODOS

Yeast cells were grown overnight in raffinose-containing media at 30°C, then grown for 5-6 hours in galactose-containing media before harvesting for sedimentation.

OD600 was normalized to 1.2 and 200mL cells were spun down and resuspended in 12mL lysis buffer (30mM HEPES pH 7.4, 150mM NaCl, 1% glycerol, 0.5% Triton X-100, 5mM EDTA, 1mM DTT, 1mM PMSF, 1% fungal protease inhibitor cocktail (Sigma)). The suspension was then passed three times through a French press (Emulsiflex C-3) to lyse the cells. Lysates were cleared by centrifugation (6000g for 5 minutes) and a 200ul fraction was taken to represent the total cellular protein content and boiled for five minutes in 3x SDS sample buffer. Another 200ul aliquot was separated into soluble and pelleted fractions by centrifugation at 100,000g for 15 minutes. The soluble fraction was boiled for five minutes in 3x SDS sample buffer and the pellet was dissolved and boiled for 10 minutes in 1x SDS sample buffer. 10% of the total and soluble protein fractions and 20% of the pelleted protein fraction were separated via SDS-PAGE and

immunoblotted as described above.

113 2.4.5 Fluorescence Microscopy

Yeast cultures grown overnight in raffinose media were diluted to early/mid log-phase in galactose media and grown for >5 hours at 30°C to induce hnRNP expression.

Live, untreated cells were visualized at 100x magnification on a Leica-DMIRBE microscope. To visualize nuclear material, cells were incubated for 15-30 minutes at room temperature with Hoechst 33342 stain (167µg/mL). All images were analyzed and processed using ImageJ software.

2.4.6 SDD-AGE

Yeast cells were grown overnight in raffinose-containing media, then harvested by centrifugation and resuspended in galactose-containing media to induce protein expression. Following 6-hour induction, yeast were pelleted by centrifugation, washed in water and then resuspended in spheroplasting solution (1.2 M D-sorbitol, 0.5 mM MgCl, 220 mM Tris, pH 7.5, 50 mM β-mercaptoethanol and 0.5 mg/ml Zymolyase 100T) and incubated for 1 hour at 30˚C. Spheroplasts were collected by centrifugation (500 rcf for 5 minutes) and resuspended in lysis buffer (100 mM Tris (pH 7.5), 50 mM NaCl, and 2x Sigma Protease Inhibitor cocktail—P8215). The suspensions were vortexed at high speed for 1 minute at 4˚C and then snap-frozen in dry ice and ethanol. Samples were thawed on wet ice and protein concentrations were determined by BCA Protein Assay.

4X sample buffer (2X TAE, 20% glycerol, 2 or 8% SDS, 10% β-mercaptoethanol, and bromophenol blue) was added and lysates were incubated for 5 minutes at room temperature. Samples of the same total protein concentration were loaded on a 1.5%

Agarose gel with 1X TAE and 0.1% SDS, cast in a horizontal slab electrophoresis apparatus tray. Samples were run at 5 V/cm gel length in a cold room until dye front was

114

1.5 cm from the end of the gel. The samples were then transferred overnight to a nitrocellulose membrane as previously described [358]. The membrane was processed for Western Blot analysis with GFP (Roche Cat. No. 11814460001) primary and α-Mouse (Rockland Anti-MOUSE IgG (H&L) IRDye800) secondary antibodies and proteins were detected using a Li-Cor Odyssey Model 9120.

2.4.7 Confocal microscopy

Yeast coexpressing GFP-tagged hnRNP constructs and mCherry- or RFP-tagged stress granule and P-body proteins were grown overnight at 30°C in non-inducing raffinose media lacking uracil and leucine. Cultures were spun down and cell pellets were resuspended in galactose media to induce hnRNP expression. After 5-6 hours of induction at 30°C, cells were harvested for microscopy. Live, unstained cells were imaged using a spinning disk confocal microscope equipped with a Yokogawa CSU X1 scan head combined with an Olympus IX 81 microscope. Acquisition and hardware were controlled by MetaMorph, version 7.7 (Molecular Devices, Downingtown PA). An Andor iXon3 897 EMCCD camera (Andor Technology, South Windsor CT) was used for image capture. Solid-state lasers for excitation (488 nm for GFP, 561 nm for RFP/mCherry) were housed in a launch constructed by Spectral Applied Research (Richmond Hill, Ontario, Canada). All images were analyzed and processed using ImageJ software.

2.4.8 Protein Purification

WT and mutant hnRNPA1 and hnRNPA2 were expressed and purified from E.

coli as GST-tagged proteins. Expression constructs were generated in pDuet to contain a TEV-cleavable site, resulting in a GST-TEV-hnRNP construct. GST-TEV-hnRNP was

115

overexpressed in E. coli BL21-CodonPlus(DE3)-RIL cells (Agilent) and purified under native conditions using a glutathione-sepharose column (GE) according to the

manufacturer’s instructions. Proteins were eluted from the glutathione sepharose with assembly buffer (hnRNPA1 D262V 105-320: 50 mM Tris-HCl, pH 8, 200 mM trehalose, and 20 mM glutathione; all other constructs: 40mM HEPES-NaOH, 150mM KCl, 5%

glycerol, 20mM glutathione, pH 7.4). Protein was centrifuged for 10 min at 16,100g, and supernatant was separated from pellet to remove any protein aggregates. Protein concentration was determined by Bradford assay (Bio-Rad) in comparison to BSA standards.

2.4.9 Sedimentation analysis of hnRNPA1 fibrillization

To follow the reaction kinetics by sedimentation analysis, at different time points, samples were centrifuged at 16,100 g for 10 min at 4°C. Supernatant and pellet fractions were then resolved by SDS-PAGE and stained with Coomassie Brilliant Blue. The amount of protein in either fraction was determined by densitometry in comparison to known quantities of hnRNPA1/hnRNPA2.

2.4.10 Transmission electron microscopy

Samples (10 µl) were adsorbed onto glow-discharged 300-mesh

Formvar/carboncoated copper grid (Electron Microscopy Sciences) and stained with 2%

(w/v) aqueous uranyl acetate. Excess liquid was removed, and grids were allowed to air dry. Samples were viewed by a JEOL 1010 transmission electron microscope.

2.4.11 ThT fluorescence

ThT fluorescence was used to assess fibrillization as previously described [359].

116

2.4.12 Genetic deletion screen for toxicity modifiers

We used the synthetic genetic array (SGA) technique to screen the collection of non-essential only yeast knockout strains. Screens were performed as described [360-362] with some modifications [363], using a Singer RoToR HDA (Singer Instruments, Somerset, UK). The galactose-inducible expression constructs (pAG416Gal-hnRNPA1 and pAG416Gal-hnRNPA2B1) were introduced into MATα strain Y7092 to generate the query strains. Query strains were mated to the yeast haploid deletion collection of non- essential genes (MATa, each gene deleted with KanMX cassette). Diploids were selected for by plating yeast onto glucose media lacking uracil with G418 added.

Diploids were then grown on 2% YPD prior to sporulation. Yeast were induced to undergo sporulation by plating on media containing 1.5% potassium acetate, 0.1%

glucose, 0.25% yeast extract, 0.01% amino-acids supplement mixture (2 g histidine, 10 g leucine, 2 g lysine, 2 g uracil), and 50 mg/L G418. Glucose media lacking histidine, arginine, lysine, and uracil and supplemented with canavanine and thialysine was used to select for MATa haploids. G418 was then used to select for MATa haploids harboring the KanMX cassette and yeast were grown in the presence of glucose (hnRNPA1 or hnRNPA2 expression ‘‘off’’) or a 1:1 mixture of sucrose and galactose (hnRNPA1 or hnRNPA2 expression ‘‘on’’). After growth at 30ºC for 2 days for glucose plates and 4 days for sucrose/galactose plates, plates were photographed and colony sizes

measured by ImageJ image analysis software, based on Collins et al. 2006 [364]. The screen was repeated twice and hits were selected and validated by repeat

transformations and spotting on 1:1 or 3:1 sucrose:galactose.

117

CHAPTER 3: CONCLUSIONS AND FUTURE DIRECTIONS

Using Saccharomyces cerevisiae as a model system, we have mapped the determinants of protein toxicity for two RBPs with PrLDs that are known to cause MSP, a degenerative syndrome encompassing features of ALS, FTD, IBM, and PDB. We have established that both hnRNPA1 and hnRNPA2 require an intact RRM and a portion of the PrLD, and thus propose a mechanism of toxicity that requires both RNA-binding and the formation of protein assemblies. It is unknown whether the toxic species in yeast is a solid fibrous state, a liquid droplet-like assembly, a soluble oligomer or a combination of these. We have highlighted the importance of the splicing machinery, the yeast protein quality control machinery, and RNP-granule components to the toxicity of both hnRNPA1 and hnRNPA2 through a genetic deletion screen to identify modifiers of toxicity.

Identifying cellular pathways that affect RBP-mediated toxicity can give us insight into the mechanisms underlying cell death. Importantly, all suppressors of hnRNPA1 also suppressed hnRNPA2 toxicity, and vice versa, suggesting a shared molecular pathway underlying the pathogenesis of disease caused by these proteins. It was also notable that there was very little overlap between modifiers of hnRNPA1 and hnRNPA2 toxicity and modifiers of other toxic, disease causing RBPs, specifically TDP-43 and FUS. Our data suggest mechanistic differences underpinning toxicity among these structurally and functionally similar proteins.

We have also expanded the repertoire of disease substrates rescued by

engineered variants of the protein disaggregase Hsp104 (A503S, V426L, and A437W).

These potentiated disaggregases with elevated ATPase activity robustly suppress the

118

toxicity of TDP-43, FUS, and α-synuclein in yeast, whereas Hsp104 has no effect

[234,235]. We have shown that these Hsp104 mutants, and likely many others, given the wide genetic landscape that supports rescue of cell viability in the setting of TDP-43 and FUS expression [234,235], are also able to suppress hnRNPA1 and hnRNPA2 toxicity.

Ultimately, adaptation of Hsp104 for use in patients with neurodegenerative diseases could represent a broadly applicable therapeutic strategy that does not require the use of genotyping or biopsy to identify patient-specific protein pathology.

These studies will provide a framework for extending the study of MSP to mammalian and other higher order systems, and future work will use the insights gleaned from these experiments to explore candidate therapeutic agents in Drosophila, mouse models, or primary neuronal cultures. There remain, however, unanswered questions and further yeast studies that could potentially add to our knowledge base and generate additional leads for exploration in other models. A genetic overexpression screen to complement our deletion screen would be a natural first step, and a high-throughput small molecule screen could tease out additional information about specific pathways affected by hnRNPA1 and hnRNPA2. The results from our deletion screen provide an arsenal of genes and pathways that can be targeted in a host of disease models using short interfering RNAs (siRNAs), antisense oligonucleotides (ASOs), or CRISPR/Cas9 technology [365-367]. They also raise a number of mechanistic questions that could begin to be answered in yeast.

It was particularly curious to us that deletion of DBR1, which encodes the lariat-debranching enzyme, did not suppress the toxic effects of hnRNPA1 and hnRNPA2, as it does for TDP-43 and FUS. Loss of Dbr1 causes the cytoplasmic accumulation of

119

intronic RNA lariats as they are excised from mRNA transcripts during splicing and cannot be degraded in the absence of the initial debranching step [251,368]. It is proposed that in the absence of lariat debranching, the undegraded RNA lariats bind to FUS and TDP-43, acting as ‘decoy’ interactors and preventing these toxic proteins from sequestering essential RNA and proteins [251]. We hypothesized that this mechanism would be widely applicable across toxic RBPs, however the observation that hnRNPA1 and hnRNPA2 toxicity was unaffected by DBR1 deletion led us to consider the possibility that the nucleotide content of the yeast intronic genome is preferentially bound by TDP-43 and FUS, but not hnRNPA1 or hnRNPA2. We propose that noncoding RNAs tailored to the preferred binding motifs of hnRNPA1 or hnRNPA2 and expressed at high levels in the cytoplasm could functionally replace RNA lariats and buffer protein toxicity as non-essential binding partners. One approach to achieving this goal would be to design long, repetitive RNA sequences that could be expressed as circular RNA. Circular RNA is, in general, protected from cytoplasmic degradation, likely because, without linear ends, it is not recognized by the RNA-decay machinery [369]. Delivery of exogenous RNAs,

though not without difficulties, is currently being explored in various forms, including siRNAs and ASOs, for a range of disorders including hepatitis B, cardio-metabolic disorders, and rare genetic disorders [367]. Three ASO therapeutics have been approved for use by the FDA [367]. Though these therapies are aimed at silencing endogenous RNA, the same principles of delivery could be applied to introduce a circular non-coding RNA species to buffer the toxicity of neurodegeneration-causing RBPs.

120

We are especially intrigued by the ability of either LSM6 or LSM7 deletion to suppress the toxicity of hnRNPA1 and hnRNPA2, because the encoded proteins both participate in two disparate aspects of RNA metabolism: splicing and degradation [40]. It would be of great utility to discern which of these processes, when perturbed, disrupts the toxicity of hnRNPA1 and hnRNPA2. Lsm6 and Lsm7 assemble into two distinct heteroheptameric complexes; Lsm1-7 is localized to P bodies in the cytoplasm, sites of mRNA degradation, and Lsm2-8 is a component of the spliceosomal U6 snRNP

[202,203]. Of the eight proteins present in these two complexes, three are non-essential (Lsm1, Lsm6, and Lsm7) [202], and, therefore, candidate toxicity suppressors within the parameters of our deletion screen. Lsm1-7 play a role in promoting decapping in the mRNA degradation pathway, as does another deletion suppressor that emerged from our screen, Sbp1 [200,201,203]. Our evidence suggests that it is less likely that the decapping role of Lsm6 and Lsm7 is crucial to the toxicity of hnRNPA1 and hnRNPA2 for a number of reasons. First, the third non-essential gene in the Lsm1-7 complex was not found to be a suppressor of hnRNPA1 or hnRNPA2. Additionally, we did not find deletion of other elements of the decapping machinery, including Pat1 and Dhh1 [177], to have any effect on hnRNPA1- or hnRNPA2-mediated toxicity. Moreover, Sbp1 is also found in stress granules, suggesting that its role in hnRNPA1 and hnRNPA2 toxicity may be related to a process other than RNA decapping and decay [200].

An alternative hypothesis is that splicing perturbations from the disruption of the Lsm2-8 complex inhibit the toxic effects of hnRNPA1 and hnRNPA2. This hypothesis could be confirmed by treating cells with a small molecule inhibitor of splicing concurrent with overexpression of hnRNPA1 or hnRNPA2. If, in fact, the inhibition of splicing

121

suppresses the toxicity of these RBPs, it could be due to the effective depletion of spliced yeast genes. In yeast, intron-retaining transcripts are degraded either by the nuclear exosome or by NMD and do not get translated [349]. Experimentally determining whether the depletion of specific essential proteins through reduced splicing efficiency reduces hnRNPA1 and hnRNPA2 toxicity could yield insight into additional toxicity-modifying biologic pathways that were not revealed by scanning the non-essential genome. Just under 5% of the yeast genome (283 genes) contains introns [370]. After filtering this list to exclude non-essential genes (which were explored in our deletion screen), each remaining yeast gene with an intronic sequence could be tested for modifying effects on hnRNPA1 and hnRNPA2 toxicity. This could be done in two ways.

Knowing that we expect decreased gene expression to suppress toxicity, we could overexpress each gene and look for those that enhance toxicity in the setting of hnRNPA1 or hnRNPA2 overexpression. This approach could prove difficult because hnRNPA1 and hnRNPA2 both confer a high level of toxicity at baseline. An alternative approach would include generating temperature-sensitive, conditional mutant strains for each intron-containing gene. Temperature-sensitive strains typically allow for the

manipulation of gene expression levels within a range of growth temperatures [371].

Altering growth conditions to reduce expression of individual essential genes could reveal genetic interactions that modify RBP toxicity. This approach could be expanded to include inhibition of all untested essential yeast genes, but the information from our deletion screen suggests that intron-containing genes may be a fruitful starting point.

An important question that has not been resolved is: what species represents the toxic conformation of hnRNPA1 and hnRNPA2 in S. cerevisiae? A portion of the PrLD is

122

required for maximal toxicity, suggesting that the formation of protein assemblies is crucial for toxicity. Whether these toxic assemblies are liquid droplets, fibrillar aggregates, or perhaps prefibrillar oligomers remains unclear. Several pieces of

evidence suggest that, in yeast, more stable fibrillar hnRNPA1 or hnRNPA2 aggregates may be protective, just as more stable aggregates are the benign species in a yeast model of Htt toxicity [353]. First, hnRNPA1 lacking a crucial steric zipper motif does not readily fibrillize in vitro, but does form liquid droplets [134,222]. This construct is highly toxic in yeast (Figure 8). Moreover, the PrLD of hnRNPA1^D262V forms stable aggregates that are detergent-insoluble in yeast, and this construct is benign (Figures 8 and 10).

Finally, the human homologue of Hsc82 negatively regulates amyloidogenesis [351].

Deletion of Hsc82, which suppresses hnRNPA1 and hnRNPA2 toxicity in yeast, may shift the relative abundance of protein conformers away from oligomers and droplets and towards stable cross-β structures. It would be useful to examine whether toxic hnRNPA1 and hnRNPA2 form liquid droplets in yeast, and whether the elimination of these species can reduce toxicity.

Aliphatic alcohols, including 1,6-hexanediol, 2,5-hexanediol, 1,5-pentanediol, or 1,4-butanediol, can disrupt the weak interactions between LCDs that mediate LLPS, and they have been proposed as a method of elucidating whether cellular structures form via LLPS [217,229,230]. New evidence suggests, however, that 1,6-hexanediol can also disrupt elements of cytoskeletal organization and reduce cellular viability [372]. It would, therefore, not be straightforward to examine which, if any, hnRNPA1 and hnRNPA2 constructs form liquid-like assemblies simply by exposing cells to aliphatic alcohols and

123

assessing for alterations in hnRNPA1 or hnRNPA2 localization or dissolution of hnRNPA1 or hnRNPA2 foci using fluorescence microscopy.

Finally, the contribution of hnRNPA1 and hnRNPA2 mutations to the overall landscape of neurodegeneration is currently unknown in that we do not yet know how frequently these mutations occur or how penetrant they are. The discovery of hnRNPA1 and hnRNPA2 mutations in MSP was rapidly followed by the identification of additional hnRNPA1 and hnRNPA2 mutations in patients with sporadic and familial ALS [94,134], and we expect the number of patients suffering from neurodegenerative phenotypes with identified mutations in hnRNPA1 or hnRNPA2 to grow as our knowledge of disease increases. Moreover, mutations in the PrLD of hnRNPDL, leading to D378N or D378H substitutions, have now been linked to limb-girdle muscular dystrophy type 1G [373]. We anticipate that over time additional RBPs with PrLDs will continue to emerge in

connection with degenerative diseases [31,41].

124 BIBLIOGRAPHY

1. Khurana V, Lindquist S (2010) Modelling neurodegeneration in Saccharomyces cerevisiae: why cook with baker's yeast? Nat Rev Neurosci 11: 436-449.

2. Lepesant J-A (2015) The promises of neurodegenerative disease modeling. Comptes Rendus Biologies 338: 584-592.

3. Swinnen B, Robberecht W (2014) The phenotypic variability of amyotrophic lateral sclerosis. Nat Rev Neurol 10: 661-670.

4. Li YQ, Tan MS, Yu JT, Tan L (2016) Frontotemporal lobar degeneration: mechanisms and therapeutic strategies. Mol Neurobiol 53: 6091-6105.

5. Ferrari R, Kapogiannis D, Huey ED, Momeni P (2011) FTD and ALS: a tale of two diseases. Curr Alzheimer Res 8: 273-294.

6. Bunner KD, Rebec GV (2016) Corticostriatal dysfunction in Huntington's disease: the basics. Front Hum Neurosci 10: 317.

7. Gitler AD (2008) Beer and bread to brains and beyond: can yeast cells teach us about neurodegenerative disease? Neurosignals 16: 52-62.

8. Lo Bianco C, Shorter J, Regulier E, Lashuel H, Iwatsubo T, et al. (2008) Hsp104 antagonizes alpha-synuclein aggregation and reduces dopaminergic

degeneration in a rat model of Parkinson disease. J Clin Invest 118: 3087-3097.

9. Skovronsky DM, Lee VM, Trojanowski JQ (2006) Neurodegenerative diseases: new concepts of pathogenesis and their therapeutic implications. Annu Rev Pathol Mech Dis 1: 151-170.

10. Liscic RM, Grinberg LT, Zidar J, Gitcho MA, Cairns NJ (2008) ALS and FTLD: two faces of TDP-43 proteinopathy. Eur J Neurol 15: 772-780.

11. Mathis S, Couratier P, Julian A, Vallat J-M, Corcia P, et al. (2016) Management and therapeutic perspectives in amyotrophic lateral sclerosis. Expert Review of Neurotherapeutics: 1-14.

12. Cummings J, Aisen PS, DuBois B, Frölich L, Jack CR, et al. (2016) Drug

development in Alzheimer’s disease: the path to 2025. Alzheimer's Research &

Therapy 8: 39.

13. Farzanehfar P (2016) Towards a better treatment option for Parkinson's disease: a review of adult neurogenesis. Neurochem Res 41: 3161-3170.

14. Bose S, Cho J (2016) Targeting chaperones, heat shock factor-1, and unfolded

14. Bose S, Cho J (2016) Targeting chaperones, heat shock factor-1, and unfolded